Computers & Education 132 (2019) 16–30

Contents lists available at ScienceDirect

Computers & Education

journal homepage: www.elsevier.com/locate/compedu

MOOC learners’ demographics, self-regulated learning strategy,

T
perceived learning and satisfaction: A structural equation modeling
approach
Kun Li∗
Doctor of Physical Therapy, Duke University, Durham, NC, USA

ARTICLE INFO ABSTRACT

Keywords: Massive Open Online Courses (MOOCs) provide a great platform to study individual and group
MOOC differences of learners in perceptions, motivations, and behaviors under self-directed learning
Self-regulated learning context. This study examined the relationships, in particular, influential relationships, among
Perceived learning MOOC learners' demographics, their self-regulated learning (SRL) strategy usage, perceived
Satisfaction
learning, and satisfaction. Participants were 4503 learners from 17 Coursera courses who re-
Culture
sponded to an online survey in 2018. Structural equation modeling showed that participants' age,
gender, highest degree, and the number of online courses previously taken significantly predicted
both goal setting and environment structuring usage. Previous experience with the course topics
only predicted goal setting, not environment structuring. Gender, goal setting and environment
structuring strategy usage predicted participants' perceived affective learning. Highest degree,
the number of online courses previously took, goal setting, environment structuring strategy
usage and perceived affective learning predicted participants' satisfaction with the course.
Participants identified themselves with a Latin America culture had better environment struc-
turing strategy usage than any other cultural group and higher perceived affective learning than
the other cultural groups except for Other. The results provided implications for researchers
studying self-directed learning environments, differences in learning of learners with diverse
backgrounds, and SRL behaviors, as well as for educators dealing with increasing SRL strategy
usage, improving online learners’ satisfaction and teaching cross-culturally.

1. Introduction

Since the rapid expansion of online education, students' behaviors, perceptions, and motivations in online environments have
been studied extensively for both theory and practice development. Pintrich and De Groot (1990) urged more researchers to in-
vestigate the relationships between students' motivations and behaviors in different learning contexts to understand how these two
factors independently or jointly influence students' learning. Practically, designing effective and engaging learning environments
requires the knowledge of the factors that influence students' learning and perceptions. For instance, in online learning contexts,
course design, interactions with instructors, and interactions with students were three factors having an influence on students’
perceived learning and satisfaction with the online course (Swan, 2001). Practitioners can focus on enhancing the three areas when
designing online courses.

∗
2200 W Main St, Durham, NC 27705, USA.
E-mail address: kl195@duke.edu.

https://doi.org/10.1016/j.compedu.2019.01.003
Received 28 September 2018; Received in revised form 31 December 2018; Accepted 6 January 2019
Available online 11 January 2019
0360-1315/ © 2019 Elsevier Ltd. All rights reserved.
K. Li Computers & Education 132 (2019) 16–30

With the more diversified student body in modern educational contexts, individual and group differences in attitudes, behaviors,
and perceptions of online learning environments have also been examined. Gender differences in perceived learning, demographics
that associated with online students' dropout, perceived cognitive learning differences among different cultural groups are some
examples in this research area (Lee & Choi, 2011; McCroskey, Sallinen, Fayer, Richmond, & Barraclough, 1996; Rovai & Baker, 2005).
The overall findings generally support that individual/group differences exist. This information can be used to develop targeted
interventions aimed at changing online learners’ attitudes, increasing their motivations and preventing them from dropping out for a
specific group/individual learners.
Online education and global reach shorten the distance between learners from different cultures. Previous studies have reported
the differences in values, beliefs, and behaviors of students from different cultures. An example is the phrase of the “paradox of the
Chinese learners” where scholars tried to understand why Chinese learners outperformed learners from the West even though the
Chinese educational system focused too much on memorization instead of understanding (Watkins, 2000). Hofstede (1986) described
the differences of student/teacher and student/student interactions in terms of a four dimension model: Individualism versus Col-
lectivism, large versus small Power Distance, strong versus weak Uncertainty Avoidance, and Masculinity versus Femininity, which
explained the cultural differences among many countries. In addition, studying similarities and differences between different cultures
helps educators to develop cultural competency and effectively teach cross-culturally. As early as 2001, Sleeter (2001) proposed to
have large-scale curriculum preparing teachers to teach cross-culturally in the U.S. Nowadays, culturally diverse schools are growing
in not only the U.S. but many other countries. Thus, the research examines cross-culture differences in teaching and learning is
imperative.
As a special type of online education, the Massive Open Online Course (MOOC) has even more diverse learner characteristics. The
early batch of MOOC research found there was a diverse student body in terms of locations and demographics, as well as behaviors in
courses (Breslow et al., 2013; Kizilcec, Piech, & Schneider, 2013). The variety of motivations for enrolling in MOOCs is another
recognizable difference between MOOCs and traditional credit-bearing online courses. Reasons for enrolling vary from preparation
for advanced education to simply examining an online course (citation removed). Because of the wide range of learner demographics
and motivations, their behaviors can differ greatly from completing the course with a good grade to totally disengaged with the
course (Kizilcec et al., 2013). Learners’ motivations and participation in a MOOC also influence their performances greatly (De Barba,
Kennedy, & Ainley, 2016). Similarly, their self-regulated learning (SRL) levels and strategy usage can be quite different, such as that
professionals reported a higher level of SRL strategy usage than novice learners in MOOCs(Hood, Littlejohn, & Milligan, 2015). The
wide range of learners, their behaviors and motivations have made MOOCs a great platform to study individual and group differences
in teaching and learning.
Several recent attempts have been made to reveal the individual or group differences in MOOC learners' attitudes, perceptions,
and behaviors (e.g. De Barba et al. (2016); Kizilcec, Pérez-Sanagustín, and Maldonado (2017)). However, little attention has been
paid to the differences in MOOC learners' perceptions and behaviors among different cultural groups. As mentioned above, MOOC
learners can be from any one country in the world, which adds another layer of the diverse background of the learners. In addition, no
existing study has examined the relationships between the four variables: MOOC learners’ demographics, learning behaviors, par-
ticularly SRL behaviors, perceived learning and satisfaction, including mediating effects. This study seeks to explore the relationships
among the four variables by applying a structural equation model (SEM). The paper is organized as follows: section 2 reviews relevant
literature on demographics, SRL, perceived learning, and satisfaction in online and MOOC environments. Hypotheses are formed
from the demonstrated relationships and particularly those contradicted results in previous research. Section 3 describes the research
method, data collection, and data analysis. Section 4 presents the study results. Section 5 discusses the results, reveals the study
limitations, and provides implications for researchers and practitioners. Section 6 concludes the study.

2. Literature review

2.1. Student demographics and online learning

Students' demographics have long been studied for relationships with many aspects of online learning. Age, gender and prior
knowledge are among the frequently investigated demographic variables. Prior experience of online learning has been identified as
an important demographic factor that may affect students online learning experiences. Yukselturk and Bulut (2007) reported no
relationship between students' general demographic information, such as age and gender, and their success in online courses. Gender
was shown as a significant factor in predicting learners' perceived learning — female students had higher perceived learning than
their male peers (Rovai & Baker, 2005). While Willging and Johnson (2009) concluded that students' demographics did not influence
their dropout decisions from online courses, in a literature review on online course dropout, Lee and Choi (2011) identified three
groups of factors that affect students’ dropout. Student demographics, including elements like “academic background, relevant ex-
periences, relevant skills, and psychological attributes” (p. 604), was among the three factors.
There are a number of studies focusing on learners' demographics and their learning outcomes in Massive Open Online Courses
(MOOCs). MOOCs are special online environments in that the students enrolled in MOOCs can be much more diverse in demo-
graphics (Breslow et al., 2013). Thus, demographic variables, which are not reported in traditional learning environments because
they are not diversified, such as students' highest degree obtained and current job status have been introduced. Morris, Hotchkiss, and
Swinnerton (2015) identified that learners' age, prior online learning experience, educational attainment, and job status predicted
their learning outcomes. Kennedy, Coffrin, De Barba, and Corrin (2015) reported that learners' prior knowledge of the topic is the best
predictor for success. And DeBoer, Stump, Seaton, and Breslow (2013) found learners’ educational attainment a significant predictor

17
K. Li Computers & Education 132 (2019) 16–30

for success, but not their prior knowledge of the course topic. The letter H with a numeric number (e.g. H1) denote the hypothesis and
its number in this article.
H1. Gender significantly predicts perceived learning.
H2. Experience significantly predicts perceived learning.

2.2. Self-regulated learning

Self-regulated learning (SRL) strategies are defined as ”the strategies that students use to regulate their cognition as well as the use
of resource management strategies that students use to control their learning” (Pintrich, 1999, p. 459). Self-regulated learners are
aware of their strengths, weaknesses and they are able to attribute time, resources, and mental effort toward their goals in learning
(Zimmerman, 2002). Paechter, Maier, and Macher (2010) stated that SRL was positively related to learning achievements.
H3. SRL strategy usage significantly predicts perceived learning.

2.2.1. Student backgrounds and self-regulated learning

Researchers have investigated the relationship between students' backgrounds, such as demographics, and their learning strategy
usage, including SRL strategies. Some researchers reported positive relationships between certain demographics and strategy usage.
For example, Colorado and Eberle (2012) found that older and higher level (graduate) students tended to be more self-regulated in
learning. Older students used more advanced strategies to monitor their own learning behaviors and used more frequently than
younger students (Lan, 2005). Vermunt and Vermetten (2004) reported that older and more experienced learners showed better
mastery and usage of effective learning strategies. Demographic variables like gender influenced students’ cognitive strategy usage
but not regulation/management strategy usage (Wolters & Pintrich, 1998). Artino and Stephens (2009) found that graduate students
showed more critical thinking pattern during learning than undergraduate students when their experiences were controlled. Students
who had taken online courses before used more SRL strategies (Wang, Shannon, & Ross, 2013). Law, Chan, and Sachs (2008) did not
find significant different SRL strategy usage between students of different ages, but they did reveal that female students reported
more SRL strategy usage than their male peers. Wang et al. (2013) reported that students who had more successful prior online
learning experiences tended to use more SRL strategies.
In MOOCs, researchers also investigate the relationship between student demographics and self-regulated learning because
compared to traditional online courses, MOOCs require learners, even more, to be self-regulated in order to learn. As a first step to
examine SRL in MOOCs, researchers have studied the differences of self-regulated behaviors in self-reported format between student
groups. For example, Hood et al. (2015) found that learners who were pursuing a higher degree in the subject tended to be more self-
regulated in the same subject MOOC. It was assumed that the SRL strategy ability was transferable from their own learning contexts.
In addition, learners who were already professionals in the field tended to use more SRL strategy. The authors hypothesized that
learners who were more familiar with the content and more confident were better able to use SRL strategies. Similarly, Kizilcec et al.
(2017) reported that learners with Ph.D. used more SRL strategies than non-PhDs while learners who were current students used
fewer strategies than non-students. The authors also found that learners who had different motivations for enrolling in the MOOC
reported different numbers of SRL strategy usage. The authors suggested these factors be integrated when designing adaptive learning
paths for learners in MOOCs.
H4. Number of MOOCs previously taken significantly predicts SRL strategy usage.
H5. Degree significantly predicts SRL strategy usage.
H6. Experience significantly predicts SRL strategy usage.
H7. Age significantly predicts SRL strategy usage.
H8. Gender significantly predicts SRL strategy usage.

2.2.2. Culture, learning and SRL

Because of the differences in educational systems, cultural values, beliefs, and emphasis on learning, it is possible that students
from different cultural backgrounds have different perceptions towards learning. For example, Li (2005) pointed out that education
in Confucius cultures emphasized on virtual while Western education focused on the mind. These differences will influence students'
beliefs and learning behaviors. Purdie, Hattie, and Douglas (1996) reported that Japanese and Australian students had different
perceptions toward learning. Researchers have examined the relationship between culture and SRL beliefs and strategy usage.
Pintrich (2003) emphasized the importance of understanding cultural impacts on individuals’ beliefs about self, learning, motivation,
and various concepts related to SRL. Wolters, Pintrich, and Karabenick (2005) stated that SRL strategy use was very context de-
pendent, meaning, a student might use a completely different set of strategies when studying different subjects or under different
learning circumstances. They also stressed the needs to extend the SRL work to a more diverse context with different ethnic groups.
Purdie et al. (1996) reported similar SRL strategy usage between Australian and Japanese students despite their different views of
education.

18
K. Li Computers & Education 132 (2019) 16–30

Several studies found that students in different countries with different cultural beliefs used different SRL strategies. Purdie and
Hattie (1996) discovered slight differences in SRL strategy usage between Japanese and Australian students, specifically using
memories-based strategies. They also revealed that the SRL strategies Japanese students who studied in Australia were different from
their Japanese peers in Japan and Australian students, showing the SRL strategy usage could be affected by both culture and learning
contexts. Unlike the student-centered classroom emphasis in western contexts, a study conducted in Hong Kong revealed that when
teachers were involved in students’ learning, students tended to use more SRL strategies (Lee, Yin, & Zhang, 2009). Studies have
found different SRL strategy usage among female students in Singapore and America Alexander, Murphy, and Guan (1998). While
Olaussen and Bråten (1999) reported that Norwegian students reported similar SRL strategy usage as American students, they
identified that female Norwegian students used more SRL strategies than their male peers.
H9. Culture significantly predicts SRL strategy usage.
H10. Culture significantly predicts perceived learning.

2.3. Student satisfaction

Students' satisfaction in learning is important because it is often found to be positively correlated with learning outcomes. Course
design and structure is one of the factors that significantly predict students' satisfaction and/or perceived learning in online courses
(Swan, 2001). Perceived learning has also been found as a significant predictor of students' satisfaction (Eom, Wen, & Ashill, 2006).
Shen, Cho, Tsai, and Marra (2013) identified that the number of previous online courses taken, gender and grade level (under-
graduate or graduate students) significantly predicted students’ self-efficacy, which significantly predicted their satisfaction. Ke and
Kwak (2013) reported that students with higher educational degrees and minority students both felt less satisfied with online
learning. It is also found that minority students tended to be more satisfied with their interactions with teachers while less satisfied
with online learning in general (Ke & Kwak, 2013).
H11. Perceived learning significantly predicts satisfaction.
H12. Number of previous online courses taken significantly predicts satisfaction.
H13. Gender significantly predicts satisfaction.
H14. Degree significantly predicts satisfaction.

2.3.1. Student satisfaction and self-regulated learning

Researchers have realized the potential relationship between SRL and student satisfaction and several attempts have been made to
address it. Because self-regulated learners constantly monitor their progress and status in order to adjust the strategies that they will
use toward their goals, these learners are always satisfied with their learning experiences (Zimmerman, 2002). Research reported that
students who had higher metacognition, time and effort management scores — elements of SRL — had higher satisfaction with the
online course (Puzziferro, 2008). Students who used more SRL strategies tended to be more satisfied with the course (Wang et al.,
2013). On the other hand, some researchers found that students’ SRL did not predict their satisfaction (Kuo, Walker, Belland, &
Schroder, 2013) nor their achievement significantly (Cho & Heron, 2015) in online learning.
H15. SRL strategy usage significantly predicts satisfaction.

2.3.2. Student satisfaction and culture

The relationship between students' cultural background and their satisfaction with courses has been investigated. Students'
cultural backgrounds influenced their satisfaction with computer-supported learning (Zhu, 2013). Zhu (2012) found significant
differences of satisfaction with online learning between Chinese students and Flemish students. Although one cannot derive that it is
the cultural differences that lead to the different satisfaction, researchers considered culture to influence participants' perceptions of
education and learning styles by communication, value, and educational system (Morse, 2003). In addition, research suggests that
online course design should consider students' cultural characteristics to make the learning experiences satisfactory (Liu, Liu, Lee, &
Magjuka, 2010). Morse (2003) studied students' perceptions of online learning and the differences between students from different
cultures. He found that students from high context cultures, which did not require a lot of explicit information because the high cues
contexts provided, had different satisfactory opinions toward online learning's various of aspects from students from low context
cultures. Fig. 1 shows all hypotheses (denoted by letter H with a number) in the graphical representation of the SEM.
H16. Culture significantly predicts satisfaction.

3. Method

3.1. Participating MOOCs and students

Seventeen MOOCs offered by Duke University on Coursera were used in the study. Courses with more than two instructors were

19
K. Li Computers & Education 132 (2019) 16–30

Fig. 1. Research model.

Table 1
Participating MOOCs and Response Rates.
Course Name Student Number Sample Size Response Rate

Advertising and Society 5464 94 1.72%

Art of the MOOC: Activism and Social Movements 5637 135 2.39%
Behavioral Finance 21616 665 3.08%
Bioelectricity: A Quantitative Approach 3549 77 2.17%
Dog Emotion and Cognition 88650 3221 3.63%
Healthcare Innovation and Entrepreneurship 3853 72 1.87%
Image and Video Processing: From Mars to Hollywood with a Stop at the Hospital 17713 498 2.81%
Introduction to Chemistry: Reactions and Ratios 14553 381 2.62%
Introductory Human Physiology 186929 2525 1.35%
Medical Neuroscience 72321 1723 2.38%
Music as Biology: What We Like to Hear and Why 27271 653 2.39%
Sports and Society 7562 260 3.44%
The Brain and Space 17840 372 2.09%
The Challenges of Global Health 17300 416 2.40%
Think Again I: How to Understand Argument 86401 1588 1.84%
Understanding 9/11: Why 9/11 Happened & How Terrorism Affects Our World Today 1361 36 2.65%
Visual Perception and the Brain 11472 250 2.18%

excluded because too many teaching styles could potentially affect students’ learning experiences. Courses that are part of a spe-
cialization were excluded because students have to pay for specialization courses to access all materials, which would make their
experiences quite different than those with free access. If a course has subsequent courses (e.g. part 1 and part 2), only the first course
was selected to avoid the possibly large number of overlapping learners. An online survey was distributed to all active students since
the launch date in each course via the Coursera email tool. The participating MOOCs, sample sizes and response rates are shown in
Table 1.

3.2. Instruments

3.2.1. Self-regulated learning strategy usage

Self-regulated learning (SRL) strategies were measured using the Online Self-regulated Learning Questionnaire (OSLQ) designed
by Barnard, Lan, To, Paton, and Lai (2009). The OSLQ measures students’ SRL usage in blended or online environments — MOOC is a
special online learning platform. The total number of the OSLQ items is 24, which makes the instrument suitable to include in a
survey containing items measuring other variables of interest. The OSLQ uses a five-point Likert scale response ranging from Strongly
agree to Strongly disagree.

3.2.2. Satisfaction with the course

Three items were designed to measure learners’ satisfaction with the course for the reason of both suitable for the research
purposes and limiting the total number of questions in the survey. The responses are five-point Likert scales ranging from Strongly
agree to Strongly disagree. The 3 items are:1. overall I am satisfied with the course. 2. I would recommend this course to other people.
and 3. I would take another course taught by the same instructor(s) of this course.

20
K. Li Computers & Education 132 (2019) 16–30

Table 2
Demographic questions and choices.
Question Choices

How many online courses have you completed before this course?
Prior to taking this course, please identify your experience with this course
topics.
Please use the culture map to identify the culture that best matches you. (it
may not fit you in a perfect sense or you may find multiple cultures apply
to you, but please identify one which describes you the most)
What is your age?
What is your gender?
What is the highest degree or level of school you have completed? If currently
enrolled, highest degree received.
• None • Less than 5 • 5–10 • More than 10
• Ifamiliar
was new to the course content • I was familiar with some topics • I was
with most topics • I was an expert on this course content
• African-Islamic • Baltic • Catholic Europe • Confucian • English Speaking •
Latin America • Orthodox Protestant Europe • South Asia
• Under 18 years old • 18–24 years old • 25–34 years old • 35–44 years old •
45–54 years old • 55–64 years old • 65–74 years old • 75 years or older
• Male • Female • Other
• Some high school, no diploma • High school graduate, diploma or the
equivalent (for example: GED) • Some college credit, no degree • Trade/
technical/vocational training • Associate degree • Bachelor's degree •
Master's degree • Doctorate degree • Professional degree (e.g. M.D., J.D.)

3.2.3. Perceived learning

Perceived learning was measured by the Cognitive, Affective and Psychomotor (CAP) Perceived Learning Scale designed by Rovai,
Wighting, Baker, and Grooms (2009). The CAP scale was designed for both face-to-face and online learning. Since many of the
participating courses did not include psychomotor skills in their learning goals, for comparable results across courses purposes, only
the six cognitive and affective items of the instrument were included in the survey. These items use a seven-point likelihood Likert
scale ranging from Not at all to Very much so.

3.2.4. Demographic information

Table 2 displays the demographic information that was collected in the survey. The participants were asked to self identify their
major culture based on a culture map from research on the World Values Survey website. The research classified cultural beliefs into
two dimensions: traditional values versus secular-rational values and survival values versus self-expression values. Then the re-
searchers mapped all countries onto a two-dimension coordinate system and clustered similar countries together to form these
cultural groups. The map can be viewed using the following link http://www.worldvaluessurvey.org/WVSContents.jsp?CMSID=
Findings.

3.3. Procedure

The Coursera courses taught by Duke University faculty with one or two instructors were selected. After the IRB approval was
obtained, the author asked all instructors teaching these courses for permission to collect data. An Email with the survey link was
then sent to all active students in these courses with the instructor(s)’ permission in February 2018. All active students since the
launch date (vs. active students since last week, last month, etc.) were selected to avoid neglecting students who were inactive for a
certain amount of time but still would like to share experiences. The survey was open for two weeks to give participants enough time
to complete.

3.4. Data analysis

This study used structural equation modeling (SEM) to test the model. All data cleaning, visualization, and analysis were done
using R 3.4 and relevant packages. Confirmatory data analysis (CFA) and SEM modeling were performed using R package lavaan
0.6.2, which has been widely used in Psychology, Social Science and other fields. Package semTools 0.5.1 was used to report several
model fit indicator measures. After removing rows with missing values, only 23 (0.51%) respondents indicated Other in the gender
question. Thus, they were removed from the analysis. The final sample consists 4503 observations (n = 4503), of which 2609 (57.9%)
are female and 1894 (42.1%) are male. Participants who were new to content, familiar with some course topic, familiar with the most
topic and an expert in course content are 1207 (26.8%), 2300 (51.1%), 900 (20.0%), and 96 (2.13%) respectively. Participants who
took zero online courses, less than 5, 5 to 10, and more than 10 are 733 (16.3%), 2151 (47.8%), 1010 (22.4%), and 609 (13.5%)
respectively. Figs. 2–4 respectively show the numbers and percentages of participants’ degree, age, and self-identified culture.
A two-step modeling approach was adopted with a CFA followed by an SEM analysis. Because the CFA result was not satisfactory
as suggested by model fit indicators and factor loadings, an exploratory factor analysis was performed by randomly splitting the
sample into two, one of which was used to run the (EFA) and the other one was used to run CFA to cross-validate the EFA results. Only
eight items from the OSLQ and two items from the CAP instruments were selected in the final analysis because other sub-scales
contradicted with the original instrument. Since instrument validation is beyond the scope of this article, only the ones consistent
with the original instrument were included. Because the numbers of respondents with Baltic, Confucian, and Orthodox cultures had the
fewest numbers of students and the three were in close positions in the culture map, they were combined into a category named Other
in the SEM model. Originally, every other cultural group was compared with English Speaking because English Speaking consisted of the

21
K. Li Computers & Education 132 (2019) 16–30

Fig. 2. Number and percentage of degree.

Fig. 3. Number and percentage of age group.

Fig. 4. Number and percentage of culture.

22
K. Li Computers & Education 132 (2019) 16–30

Table 3
Descriptive statistics of the items in the measure.
Construct and item Mean Standard deviation Skewness Kurtosis

Goal
Goal 1 3.89 1.00 −0.79 3.33
Goal 2 3.77 1.07 −0.77 3.03
Goal 3 4.01 1.01 −1.01 3.59
Goal 4 3.79 1.09 −0.77 2.95
Environment
Env 1 4.22 1.00 −1.33 4.23
Env 2 4.32 0.91 −1.52 5.23
Env 3 4.38 0.88 −1.60 5.60
Env 4 4.23 0.94 −1.34 4.57
Perceived affective learning
Learn 1 4.33 1.44 −0.97 3.73
Learn 2 4.25 1.53 −0.90 3.41
Satisfaction
Sat 1 4.62 0.76 −2.65 11.12
Sat 2 4.61 0.78 −2.58 10.44
Sat 3 4.56 0.83 −2.31 8.76

largest proportion of the sample. But later in the analysis, Latin America showed strong differences from English Speaking in several
paths. Latin America was then made the base group to which the other groups were compared.

4. Results

4.1. Descriptive statistics

The descriptive statistics of measured variables in the model are reported in Table 3, including skewness and kurtosis which are
indicators for univariate normality. Mean scores range from 3.77 to 4.62. Standard deviation scores range from 0.76 to 1.53. All but
two items’ kurtosis scores (sat 1 and sat 2) do not fall in the acceptable ranges of normality suggested by Kline (2005) (skewness does
not exceed |3| and kurtosis does not exceed |10|). But lavaan applies the diagonally weighted least squares (DWLS) estimation method
for ordinal endogenous variables. DWLS has been proved to be robust with ordinal and non-normal data (Mindrila, 2010). The two
items were not considered problematic in the analysis.

4.2. Measurement model

The CFA results are shown in Table 4. All standardized factor loadings were greater than 0.70, between 0.76 and 0.97. The R
squared scores ranged from 0.57 to 0.93, meaning that the survey items were explained by their latent variables at a range from 57%

Table 4
Results for the measurement model.
Construct and item Standardized factor loading R2 Cronbach's alpha Composite reliability Average variance extracted
(> 0.70)a (> 0.70)a (> 0.70)a (> 0.50)a

Goal 0.91 0.87 0.71

Goal 1 0.80 0.64
Goal 2 0.85 0.72
Goal 3 0.86 0.74
Goal 4 0.87 0.76
Environment 0.90 0.88 0.71
Env 1 0.86 0.73
Env 2 0.87 0.76
Env 3 0.89 0.79
Env 4 0.76 0.57
Perceived affective learning 0.88 0.79 0.78
Learn 1 0.89 0.79
Learn 2 0.88 0.77
Satisfaction 0.95 0.92 0.86
Sat 1 0.94 0.88
Sat 2 0.97 0.93
Sat 3 0.88 0.77

a
Indicates an acceptable level of reliability or validity.

23
K. Li Computers & Education 132 (2019) 16–30

Table 5
The HTMT ratio of correlation.
Goal Environment Perceived affective learning Satisfaction

Goal –
Environment 0.60 –
Perceived affective learning 0.40 0.30 –
Satisfaction 0.35 0.35 0.55 –

Table 6
Fit indices for the measurement model.
Comparative Fit Index (CFI) Tucker-Lewis Index (TLI) RMSEAa (90% Confidence Interval) SRMRb

Model indices 0.99 0.99 0.045 (0.04, 0.05) 0.02

Recommended threshold 0.95 0.95 < 0.06 < 0.08

a
: Root Mean Square Error of Approximation.
b
: Standardized Root Mean Square Residual.

to 93%. All Cronbach's alphas were greater than 0.70 indicating good reliability of items within a construct (Cortina, 1993). The
convergent validity of the measurement model was measured by two scores: the composite reliability (CR) and the average variance
extracted (AVE) (Fornell & Larcker, 1981). CR measures item internal consistency, which is considered less biased than Cronbach's
alpha. In this study, McDonald (1999)’s omega was used as a measure for CR. All CR scores in the measurement model are greater than
0.70 indicating good internal consistency (Nunnally & Bernstein, 1994). AVE is the measure of the variance that is explained by the
constructs compared to the variance explained by measurement errors. A score greater than 0.70 is considered good while a score
greater than 0.50 is considered acceptable. Judging by CR and AVE, the convergent validity of the measurement model can be
considered adequate.
Discriminant validity was tested using the heterotrait-monotrait ratio of correlations (HTMT) (Henseler, Ringle, & Sarstedt, 2015).
Table 5 shows the results of the HTMT ratio of correlations. Values greater than 0.85 indicated lack of the discriminant validity.
Results of the measurement model indicate satisfactory discriminant validity.
Multiple model fit indices were reported for both the CFA and the SEM models in the next subsection as suggested by Kline
(2005). Since the sample size of this study was quite large (n = 4503), Chi-square ( 2 ) would be almost inevitably significant. Thus,
Chi-square was not reported. Lavaan provides robust measures for these fit indices, which accounted for unbiased standard errors.
Table 6 shows the robust fit indices of the measurement model and the recommended thresholds. Recommended thresholds were
retrieved from Hu and Bentler (1999).

4.3. Structural model

A test of the structural model showed a good model fit (CFI = 0.99, TLI = 0.99, RMSEA = 0.034, RMSEA's 90% confidence
interval is (0.032, 0.036), SRMR = 0.03). All measures were reported using lavaan's robust indices, the same as the measurement
model. Fig. 5 shows the graphical description of the results of path coefficients. The results showed that three hypotheses were not
supported by the data. The results failed to reject the null hypothesis of (1) learners' experience with the course content did not
predict their perceived learning (H2 ). (2) Learners' experience with the course content did not predict their environment structuring
strategy usage (H6b ). and (3) Learners' gender did not predict their satisfaction with the course (H13 ).
For these hypotheses that the results were able to reject, Table 7 presents the findings from this study's data and results except for
the relationships of culture (reported in the following paragraph).
Because there were seven cultural groups in the study and the other groups were compared against Latin America, the results are
presented in Table 8 as a separate section. There are several differences in goal setting strategy usage and satisfaction between some
cultural groups and Latin America, but the strongest trends lied in the facts that learners identified themselves as Latin America culture
reported more environment structuring strategy usage than any other cultural groups, as well as higher perceived learning than other
cultural groups except for Other.
The proportions of variance in all endogenous variable that are explained by all of its predictors are listed below. Only 2.90% of
the variance in goal setting could be explained by all the predictors in the model while 4.50% of the variance in environment
structuring could be explained by its predictors. 35.0% of perceived learning's variance was explained by all its predictors and 18.3%

24
K. Li Computers & Education 132 (2019) 16–30

Fig. 5. Path coefficients of the research model.

* p < . 05, ** p < . 01, *** p < . 001. a, b, c, and d see Table 7.

Table 7
Results.
Predictor Results

Gender Female students reported more goal setting (GS) and more environment structuring (ES) strategies. Female students perceived lower
learning.
Age Older students reported more GS but fewer ES strategies.
Degree Students with higher degrees reported fewer GS and fewer ES strategies, but were more satisfied with the course.
Experience Students having more experiences with the content reported more GS strategies.
Number of course Students took more online courses previously reported more GS and more ES strategies, but they were less satisfied with the course.
GS strategy Students who used more GS strategies perceived higher learning and were more satisfied with the course. GS partially mediated the
effects of gender to perceived learning, degree to satisfaction, and number of courses to satisfaction. GS was the mediator of the
relationship between experience and perceived learning.
ES strategy Students who used more ES strategies perceived higher learning and were more satisfied with the course. ES partially mediated the
effects of gender to perceived learning, degree to satisfaction, and number of courses to satisfaction.
Perceived learning Students who perceived higher learning were more satisfied with the course. Perceived learning partially mediated the effects of GS and
ES on satisfaction. Perceived learning was the mediator of the relationship between gender and satisfaction.

variance in satisfaction was explained by its predictors.

4.4. Model comparison

To compare the research model with the model of which the non-significant paths removed, the 2 (chi-square change) and the Δ
CFI were reported because of the sensitivity of chi-square to sample size (Cheung & Rensvold, 2002). Table 9 lists the results of the
model comparison. Both the 2 ( p > . 05) and the Δ CFI (Δ CFI < 0.01) suggested these two models were not significantly different.

25
K. Li Computers & Education 132 (2019) 16–30

Table 8
Cultural group comparisons with Latin America.
a: culture → goal b: culture → environment

Culture β p β p
African-Islamic .01 .52 -.05 .00**
Catholic Europe -.04 .08 -.08 .00***
English Speaking .00 .94 -.10 .00***
Protestant Europe -.05 .01** -.08 .00***
South Asia .05 .01* -.05 .00**
Other -.05 .01** -.08 .00***

c: culture → perceived learning d: culture → satisfaction

Culture β p β p
African-Islamic -.04 .03* -.02 .46
Catholic Europe -.07 .00** -.01 .74
English Speaking -.17 .00*** -.02 .36
Protestant Europe -.09 .00*** .02 .34
South Asia -.07 .00*** -.06 .00**
Other -.02 .35 -.02 .37

* p < . 05, ** p < . 01, ***. p < . 001

Table 9
Results of model comparisons.
Model 2 df CFI 2 (p-value) Δ df Δ CFI

The research model 544.42 163 0.991 – – –

The revised model 575.87 172 0.993 16.448 (.059) 9 0.002

5. Discussions

5.1. Result discussions

The purpose of this study was to examine the relationships among learners' demographics, SRL strategy usage, perceived learning,
and satisfaction in MOOC environments at the same time by implementing an SEM model. Guided by a large body of literature on this
topic, it is found that many demographic variables predict learners' SRL strategy usage — in particular, goal setting and environment
structuring strategies in this study, with the exception of prior experience predicting environment structuring strategy usage.
Perceived learning can be predicted by learners' gender, culture, and their SRL strategy usage. Satisfaction can be predicted by
learners’ highest degree, the number of online courses previously taken, SRL strategy usage and perceived learning.
It is consistent with previous research that female students often reported using more self-regulated learning strategies (Pajares,
2002). Goal setting strategies were assessed specifically by Zimmerman and Martinez-Pons (1990) and female students again reported
more usage. Besides the potential fact that females do use more SRL strategies in learning than males, Pajares (2002) proposed four
possibilities of why there is a gender difference in SRL strategy usage. They are: (1) achievement is the mediator variable. (2) Males
and females respond to self-reported measures differently. (3) Males and females have a different level of confidence in certain areas.
(4) Gender stereotype may influence how people position themselves.
Learners who obtained higher degrees tended to report less goal setting and less environment structuring strategies, which
contradicts with many previous studies. For example, Kizilcec et al. (2017) found that MOOC learners who have Ph.D.s had stronger
SRL skills compared to those non-Ph.D.s. One reason is that this study examined eight degrees while Kizilcec et al. (2017) only
compared Ph.D.s with non-Ph.D.s. Another possibility may be related to learners’ motivations for enrolling. Learners with higher
degrees might take MOOCs just to explore, or to reference specific topics. Thus, there was no need to set specific goals or to control
their learning environments versus those who really wanted to learn the entire course.
This study reveals that previous experience with the course topics predicted only goal setting strategies, not environment
structuring strategies, which partially supports previous research in that the more online courses learners had taken, the more SRL
strategies they used in learning (e.g. Wang et al. (2013)). While previous research generally reports a positive relationship between
experience and SRL strategy usage (e.g. Vermunt and Vermetten (2004)), no study has investigated the relationship between par-
ticular environment structuring strategies and experience. Further research can examine the reasons for this.
The influences of age on the two SRL strategy usage show opposite results: older students tend to use less goal setting strategies
but more environment structuring strategies. This result directly contradicts with what Kizilcec et al. (2017) found, that is, older
students reported using more goal setting strategies in MOOCs. One potential explanation for this difference may be in the study
samples: this study's participants came from around the world with multiple cultures while the participants in Kizilcec et al. (2017)
were mainly Latin American learners taking courses in Spanish.

26
K. Li Computers & Education 132 (2019) 16–30

Male learners perceived higher learning than their female peers. Possible explanations can also be in Pajares (2002)’s article as
discussed above. The usage of both SRL strategies — goal setting and environment structuring — significantly predicted learners'
perceived learning, which supports the findings in Lee and Lee (2008). The more online courses a learner took, the less satisfied he/
she is with the MOOC. Although Shen et al. (2013) found a positive relationship of the number of online courses took and satisfaction,
most research has demonstrated that learners' previous experiences of online courses, not merely the number, influence their sa-
tisfaction with the current course. The fact that this study measured perceived affective learning instead of cognitive learning may
contribute to some of the contradictory results mentioned above. Learners' experience with their previous online courses was not
examined in this study. It is possible that their previous online experience was not good, or they continued enrolling but were not able
to find a satisfactory course.
Learners with higher degrees were more satisfied with the course, which contradicted the finding by Ke and Kwak (2013).
Learners' highest degree earned and their satisfaction are more complicated in MOOCs. For example, Shapiro et al. (2017) showed
that MOOC learners with a Bachelor's degree were more positive than students with lower and more advanced degrees through a
sentiment analysis of their interview transcripts. Current literature has inconsistent results about the relationship between students'
SRL strategy usage and their satisfaction with the course. For example, Cho and Heron (2015) found no relationship. Puzziferro
(2008) reported a positive relationship between several SRL strategies, including environment structuring strategies, and their levels
of satisfaction. The current study supported that learners who reported using more SRL strategies — both of them — were more
satisfied with the course.
One uniqueness of this study was the investigation of culture on MOOC learners' SRL strategy usage, perceived learning, and
satisfaction. Learners identified themselves as having Latin America culture reported more environment structuring strategies than
any other culture. The four items measuring the environment structuring strategies were about choosing quiet, comfortable places
and times with few distractions to study. No previous study in English, as far as the author is aware, was found that highlighted the
learning habits of Latin America students or whether its education emphasizes controlling learning environments. Thus, it can only be
assumed that Latin America schools or families may be good at teaching students to choose a comfortable environment to study. It is
also possible that the Latin America participants were associated with some other variables (e.g. current job status or socioeconomic
status) that were not measured in the study, and these hidden variables contributed to the better environment structuring strategy
usage. Latin America participants also showed better perceived learning than the other cultural groups except for Other. Research has
shown that people in Latin America have a high happiness level despite the fact that many countries in Latin America have high
poverty rates and unequal incomes (Rojas, 2016, pp. 1–13). The author stated that Latin Americans generally have high affective
states and high satisfaction. This may be the reason why participants of the Latin America cultural group perceived more affective
learning than many other groups. Since the group Other consisted of Baltic, Confucian, and Orthodox, it is more complicated as why
the group of Other had similar perceived learning as Latin America instead of lower. South Asia showed lower satisfaction with the
courses than Latin America while the other groups’ satisfaction did not differ significantly from Latin America. Because of the lack of
theory and research to support this point, the differences may be attributed to certain unmeasured variables or simply by chance.

5.2. Implications

There are three major implications for researchers and practitioners in the field of self-regulated learning in online learning
environments, particularly MOOC environments. The three implications, discussed in detail below with prior research findings, are
(1) influences of SRL strategy usage on learners' perceived learning and satisfaction; (2) relationships between learners’ demographics
and SRL strategy usage; and (3) differences of SRL strategy usage, perceived learning, and satisfaction among different cultural
groups.
Firstly, the present study emphasized the importance of SRL strategy usage by revealing its relationships with perceived learning
and satisfaction. Zimmerman (2002) explained that by monitoring and reflecting on their goals and progress, learners can feel more
motivated and satisfied, and then seek opportunities to improve their learning. Schunk (1990) pointed out that learners’ satisfaction
is expected to increase after they achieve their goals by allocating effort toward these goals. Learners learn better when they con-
stantly monitor their goals and progress, reflect on their experiences, and adjust effort and learning strategies. Research has shown
that SRL strategy usage is teachable even to young students in elementary schools (Dignath, Buettner, & Langfeldt, 2008).
Zimmerman (2002) stated that SRL skills are important for life-long learners. MOOCs are a form of continuing education, life-long
learning and self-directed professional development, which requires learners to have SRL skills. In addition, the current study further
supports the positive relationship between SRL strategy usage and perceived learning as well as satisfaction. Therefore, researchers
and instructional designers should investigate activities that can teach certain SRL strategies in self-directed online learning en-
vironments. It is reported that even a small activity to help students to be aware of their learning situations can help learning
(Zimmerman, 2002). Researchers can design simple interventions such as weekly emails containing “study tips” to enable this
awareness.
Secondly, different levels of SRL strategy usage — goal setting and environment structuring in this study — differed by learners.
This is of particular importance for MOOC researchers and practitioners because of the varied backgrounds and demographics of
MOOC learners. The demand for SRL strategy training or even small activities to encourage self-regulation in online learning was
discussed in the previous section. Understanding the relationships between learners’ demographics and their SRL strategy usage helps
to teach learners SRL strategies, especially to particular groups of learners. For example, lack of time has been identified as the top
reason why learners drop out of MOOCs (Shapiro et al., 2017; Zheng, Rosson, Shih, & Carroll, 2015). Designing SRL strategy training
interventions for targeted populations who may be diagnosed at risk can prevent learners from discontinuing because of reasons such

27
K. Li Computers & Education 132 (2019) 16–30

as lack of resources or lack of time. However, since the total variances of both SRL strategy usage explained by all predictors were
small (2.9% and 4.5% respectively), future studies should examine other variables that may contribute to the differences in SRL
strategy usage.
Thirdly, the differences in SRL strategy usage, perceived learning and satisfaction among learners from diverse cultures had
several implications for researchers, teachers, and instructional designers who need to design instructions cross-culturally. The
present study supports that differences among cultural groups in beliefs and behaviors exist in MOOCs. There may be the problem of
overgeneralization if the attention has been on cultural differences as a static trait (Gutiérrez & Rogoff, 2003), but like all other
literature on group differences, cultural differences bring one more dimension to individual/group differences on teaching and
learning. Researchers can examine whether and, if it is, why learners from Latin America culture had significantly higher environment
structuring strategy than any other cultural group. If there are particular reasons in Latin America educational systems or family
values that help students adopt more environment structuring strategy, this can potentially be adopted and emulated in other cultures
that may lack environment structuring guidance. Similarly, the reasons why learners with Latin America culture had higher perceived
affective learning can be examined for educational implications. In addition, teaching cross-culturally is an important component in
the educational literature. Practitioners and educators should be sensitive about teaching cross-culturally. In an interview with
instructional design professionals, Rogers, Graham, and Mayes (2007) reported that these professionals were aware of cultural
difference to some extent but they also admitted that more awareness needed to be developed. Gay (2002) argued that because
culture affects students’ attitudes, values, and behaviors, it is crucial to design culturally-relevant curricula, teaching and commu-
nication methods, and learning community. Gopal (2011) proposed that faculty in higher education need to work on their attitudes,
knowledge, and skills to develop cultural competency in order to better teach cross-culturally.

5.3. Limitations

The present study has four main limitations that, although common in this type of work, should be noted when drawing con-
clusions from the findings. Survey research that uses convenient samples lacks external validity compared to random samples.
MOOCs offered by one university on one platform were not representative of all MOOCs, and respondents to the survey in this study
were not a representative sample of all enrolled students. Although the courses selected in this study included multiple topics
compared to many research that used students from only one course, conclusions cannot be drawn to the entire MOOC students
population. The second limitation is that this study is based on learners' self-reported data. As much as self-reported data can broaden
the areas that direct observations or other achievement assessments cannot measure, problems with self-reported data have been
discussed intensely in research, including over- and under-reporting and social desirability bias (Gonyea, 2005). Future studies
should include more objective measures such as learning management system log data and test scores to offer more holistic insight
into learners’ behaviors, learning achievement and motivation. Because of the low response rate of surveys, nonresponse bias is the
third limitation. Most MOOC research using survey method suffer from this limitation. The last limitation lies in the construct
validity. Even though the study used two existing instruments measuring SRL strategy usage and perceived learning, factor analysis
did not reveal satisfactory results in terms of their sub-scales. Exploratory factor analysis results were used in this study, which left
out some items in the original instrument. Several reliability indicators suggested good reliability of the items that were used in the
study; however, future research is still needed to validate the instruments in MOOC environments.

6. Conclusion

Self-initiated and self-directed learning has become increasingly popular and important in education and in professional devel-
opment. According to one of Pew Research Center's 2016 reports, 74% of American adults had participated in some types of personal
voluntary learning activities to enrich their knowledge (source: http://www.pewinternet.org/2016/03/22/lifelong-learning-and-
technology/). Having SRL skills is essential to students' success in such self-directed learning environments, an example of which is
the MOOC. Studies on MOOC learners' SRL skills, such as the current one, emphasize the importance of SRL strategy usage by
unraveling its relationships with other crucial factors in learning. These findings suggest positive relationships between SRL strategy
usage and perceived learning, and SRL strategy usage and satisfaction. Practically, educators should emphasize the importance of
using SRL strategies in self-directed learning and create opportunities to help learners increase SRL awareness and SRL skills in
learning.
Today's student body is more and more diverse because of both globalization and online technologies. Studies have already
demonstrated that students learn differently, but personalized learning remains a challenge in education. A first step to personalized
learning can be to uncover the differences in attitudes, behaviors, and motivations by groups. MOOC learners possibly have the most
diversified backgrounds and experiences among all learning contexts, which makes the MOOC an exceptional platform to study group
differences. Results of this study show that some differences exist in learning strategy usage, perceptions and satisfaction by groups,
such as age groups and highest degrees obtained. Practically, learning activity and assessment's variability is necessary for students
with different backgrounds and motivations. Learning analytics algorithms could take these variables as input to predict at-risk
students, best learning routes, or pertinent help at an appropriate timing.
Keengwe (2010) stated that “… raising personal awareness about different cultural categories of individual differences, and how
these differences enhance or hinder the ways students and teachers generally interact with each other.” (p. 203). Theoretically,
studying the differences in learning behaviors, attitudes, and perceptions among MOOC learners from multiple cultures can con-
tribute to the cross-cultural education literature, which then adds to the broader area of cultural differences. The study results suggest

28
K. Li Computers & Education 132 (2019) 16–30

that learners identified themselves as Latin American showed higher environment structuring strategy usage and perceived learning
than many other cultural groups. Practically, revealing these cultural differences in self-directed learning explains the differences in
thinking, behavior, motivation, and choice during the learning process. Culture can be treated as a variable to predict students’
success in order to personalize learning or provide help.

Conflicts of interest

The author declares no conflict of interests.

Acknowledgement

The author would like to thank all instructors teaching the participating Coursera courses to agree on data collections in their
courses. Special thanks go to Dr. Leonard White for his careful review and suggestions on the online survey. The author would like to
thank Dr. Kimberly Manturuk for her help at the data collection stage, and Justin Johnsen for his help with the Coursera platform.
Stephanie Halpin is thanked for providing constructive feedback and editing the manuscript.

References

Alexander, P. A., Murphy, P. K., & Guan, J. (1998). The learning and study strategies of highly able female students in Singapore. Educational Psychology, 18, 391–407.
https://doi.org/10.1080/0144341980180403.
Artino, A. R., & Ttephens, J. M. (2009). Academic motivation and self-regulation: A comparative analysis of undergraduate and graduate students learning online. The
Internet and Higher Education, 12, 146–151. https://doi.org/10.1016/j.iheduc.2009.02.001.
Barnard, L., Lan, W. Y., To, Y. M., Paton, V. O., & Lai, S.-L. (2009). Measuring self-regulation in online and blended learning environments. The Internet and Higher
Education, 12, 1–6. https://doi.org/10.1016/j.iheduc.2008.10.005.
Breslow, L., Pritchard, D. E., DeBoer, J., Stump, G. S., Ho, A. D., & Seaton, D. T. (2013). Studying learning in the worldwide classroom: Research into edx's first mooc.
Research & Practice in Assessment, 8, 13–25.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9, 233–255. https://doi.
org/10.1207/S15328007SEM0902_5.
Cho, M.-H., & Heron, M. L. (2015). Self-regulated learning: The role of motivation, emotion, and use of learning strategies in students' learning experiences in a self-
paced online mathematics course. Distance Education, 36, 80–99. https://doi.org/10.1080/01587919.2015.1019963.
Colorado, J. T., & Eberle, J. (2012). Student demographics and success in online learning environments. Emporia State Research Studies, 46, 4–10.
Cortina, J. M. (1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78, 98–104.
De Barba, P., Kennedy, G. E., & Ainley, M. (2016). The role of students' motivation and participation in predicting performance in a mooc. Journal of Computer Assisted
Learning, 32, 218–231. https://doi.org/10.1111/jcal.12130.
DeBoer, J., Stump, G. S., Seaton, D., & Breslow, L. (2013). Diversity in mooc students' backgrounds and behaviors in relationship to performance in 6.002 x. Proceedings
of the sixth learning international networks consortium conference. Cambridge, MA.
Dignath, C., Buettner, G., & Langfeldt, H.-P. (2008). How can primary school students learn self-regulated learning strategies most effectively?: A meta-analysis on self-
regulation training programmes. Educational Research Review, 3, 101–129. https://doi.org/10.1016/j.edurev.2008.02.003.
Eom, S. B., Wen, H. J., & Ashill, N. (2006). The determinants of students' perceived learning outcomes and satisfaction in university online education: An empirical
investigation. Decision Sciences Journal of Innovative Education, 4, 215–235. https://doi.org/10.1111/j.1540-4609.2006.00114.x.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 39–50.
https://doi.org/10.2307/3151312.
Gay, G. (2002). Preparing for culturally responsive teaching. Journal of Teacher Education, 53, 106–116.
Gonyea, R. M. (2005). Self-reported data in institutional research: Review and recommendations. New Directions for Institutional Research, 73–89. https://doi.org/10.
1002/ir.156 2005.
Gopal, A. (2011). Internationalization of higher education: Preparing faculty to teach cross-culturally. International Journal of Teaching and Learning in Higher Education,
23, 373–381.
Gutiérrez, K. D., & Rogoff, B. (2003). Cultural ways of learning: Individual traits or repertoires of practice. Educational Researcher, 32, 19–25. https://doi.org/10.3102/
0013189X032005019.
Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. Journal of the
Academy of Marketing Science, 43, 115–135.
Hofstede, G. (1986). Cultural differences in teaching and learning. International Journal of Intercultural Relations, 10, 301–320. https://doi.org/10.1016/0147-1767(86)
90015-5.
Hood, N., Littlejohn, A., & Milligan, C. (2015). Context counts: How learners' contexts influence learning in a mooc. Computers & Education, 91, 83–91. https://doi.org/
10.1016/j.compedu.2015.10.019.
Hu, L.-t., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation
Modeling: A multidisciplinary journal, 6, 1–55.
Keengwe, J. (2010). Fostering cross cultural competence in preservice teachers through multicultural education experiences. Early Childhood Education Journal, 38,
197–204. https://doi.org/10.1007/s10643-010-0401-5.
Ke, F., & Kwak, D. (2013). Online learning across ethnicity and age: A study on learning interaction participation, perception, and learning satisfaction. Computers &
Education, 61, 43–51. https://doi.org/10.1016/j.compedu.2012.09.003.
Kennedy, G., Coffrin, C., De Barba, P., & Corrin, L. (2015). Predicting success: How learners' prior knowledge, skills and activities predict mooc performance.
Proceedings of the fifth international conference on learning analytics and knowledge (pp. 136–140). New York: ACM.
Kizilcec, R. F., Pérez-Sanagustín, M., & Maldonado, J. J. (2017). Self-regulated learning strategies predict learner behavior and goal attainment in massive open online
courses. Computers & Education, 104, 18–33. https://doi.org/10.1016/j.compedu.2016.10.001.
Kizilcec, R. F., Piech, C., & Schneider, E. (2013). Deconstructing disengagement: Analyzing learner subpopulations in massive open online courses. Proceedings of the
third international conference on learning analytics and knowledge (pp. 170–179). ACM. https://doi.org/10.1145/2460296.2460330.
Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York: Guilford publications.
Kuo, Y.-C., Walker, A. E., Belland, B. R., & Schroder, K. E. (2013). A predictive study of student satisfaction in online education programs. International Review of
Research in Open and Distance Learning, 14, 16–39. https://doi.org/10.19173/irrodl.v14i1.1338.
Lan, W. (2005). Self-monitoring and its relationship with educational level and task importance. Educational Psychology, 25, 109–127. https://doi.org/10.1080/
0144341042000294921.
Law, Y.-k., Chan, C. K., & Sachs, J. (2008). Beliefs about learning, self-regulated strategies and text comprehension among Chinese children. British Journal of
Educational Psychology, 78, 51–73. https://doi.org/10.1348/000709907X179812.

29
K. Li Computers & Education 132 (2019) 16–30

Lee, Y., & Choi, J. (2011). A review of online course dropout research: Implications for practice and future research. Educational Technology Research & Development, 59,
593–618. https://doi.org/10.1007/s11423-010-9177-y.
Lee, J.-K., & Lee, W.-K. (2008). The relationship of e-learner’s self-regulatory efficacy and perception of e-learning environmental quality. Computers in Human
Behavior, 24, 32–47. https://doi.org/10.1016/j.chb.2006.12.001.
Lee, J. C.-K., Yin, H., & Zhang, Z. (2009). Exploring the influence of the classroom environment on students' motivation and self-regulated learning in Hong Kong. The
Asia-Pacific Education Researcher, 18, 219–232.
Li, J. (2005). Mind or virtue: Western and Chinese beliefs about learning. Current Directions in Psychological Science, 14, 190–194. https://doi.org/10.1111/j.0963-
7214.2005.00362.x.
Liu, X., Liu, S., Lee, S., & Magjuka, R. J. (2010). Cultural differences in online learning: International student perceptions. Educational Technology & Society, 13,
177–188.
McCroskey, J. C., Sallinen, A., Fayer, J. M., Richmond, V. P., & Barraclough, R. A. (1996). Nonverbal immediacy and cognitive learning: A cross-cultural investigation.
Communication Education, 45, 200–211. https://doi.org/10.1080/03634529609379049.
McDonald, R. (1999). Test theory: A unified treatment. Mahwah, NJ: Erlbaum.
Mindrila, D. (2010). Maximum likelihood (ml) and diagonally weighted least squares (dwls) estimation procedures: A comparison of estimation bias with ordinal and
multivariate non-normal data. International Journal of Digital Society, 1, 60–66. https://doi.org/10.20533/ijds.2040.2570.2010.0010.
Morris, N. P., Hotchkiss, S., & Swinnerton, B. (2015). Can demographic information predict mooc learner outcomes. Proceedings of the EMOOC stakeholder summit (pp.
199–207). .
Morse, K. (2003). Does one size fit all? Exploring asynchronous learning in a multicultural environment. Journal of Asynchronous Learning Networks, 7, 37–55.
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory, Vol. 3. New York: McGraw-Hill.
Olaussen, B. S., & Bråten, I. (1999). Students' use of strategies for self-regulated learning: Cross-cultural perspectives. Scandinavian Journal of Educational Research, 43,
409–432. https://doi.org/10.1080/0031383990430405.
Paechter, M., Maier, B., & Macher, D. (2010). Students' expectations of, and experiences in e-learning: Their relation to learning achievements and course satisfaction.
Computers & Education, 54, 222–229. https://doi.org/10.1016/j.compedu.2009.08.005.
Pajares, F. (2002). Gender and perceived self-efficacy in self-regulated learning. Theory and Practice, 41, 116–125. https://doi.org/10.1207/s15430421tip4102_8.
Pintrich, P. R. (1999). The role of motivation in promoting and sustaining self-regulated learning. International Journal of Educational Research, 31, 459–470. https://
doi.org/10.1016/S0883-0355(99)00015-4.
Pintrich, P. R. (2003). A motivational science perspective on the role of student motivation in learning and teaching contexts. Journal of Educational Psychology, 95,
667.
Pintrich, P. R., & De Groot, E. V. (1990). Motivational and self-regulated learning components of classroom academic performance. Journal of Educational Psychology,
82, 33. https://doi.org/10.1037/0022-0663.82.1.33.
Purdie, N., & Hattie, J. (1996). Cultural differences in the use of strategies for self-regulated learning. American Educational Research Journal, 33, 845–871. https://doi.
org/10.3102/00028312033004845.
Purdie, N., Hattie, J., & Douglas, G. (1996). Student conceptions of learning and their use of self-regulated learning strategies: A cross-cultural comparison. Journal of
Educational Psychology, 88, 87–100. https://doi.org/10.1037/0022-0663.88.1.87.
Puzziferro, M. (2008). Online technologies self-efficacy and self-regulated learning as predictors of final grade and satisfaction in college-level online courses. American
Journal of Distance Education, 22, 72–89. https://doi.org/10.1080/08923640802039024.
Rogers, P. C., Graham, C. R., & Mayes, C. T. (2007). Cultural competence and instructional design: Exploration research into the delivery of online instruction cross-
culturally. Educational Technology Research & Development, 55, 197–217. https://doi.org/10.1007/s11423-007-9033-x.
Rojas, M. (2016). Handbook of happiness research in Latin america. chapter Happiness, Research, and Latin America. The Netherlands: Springerhttps://doi.org/10.1007/
978-94-017-7203-7.
Rovai, A. P., & Baker, J. D. (2005). Gender differences in online learning: Sense of community, perceived learning, and interpersonal interactions. Quarterly Review of
Distance Education, 6, 31.
Rovai, A. P., Wighting, M. J., Baker, J. D., & Grooms, L. D. (2009). Development of an instrument to measure perceived cognitive, affective, and psychomotor learning
in traditional and virtual classroom higher education settings. The Internet and Higher Education, 12, 7–13. https://doi.org/10.1016/j.iheduc.2008.10.002.
Schunk, D. H. (1990). Goal setting and self-efficacy during self-regulated learning. Educational Psychologist, 25, 71–86. https://doi.org/10.1207/s15326985ep2501_6.
Shapiro, H. B., Lee, C. H., Roth, N. E. W., Li, K., Çetinkaya-Rundel, M., & Canelas, D. A. (2017). Understanding the massive open online course (mooc) student
experience: An examination of attitudes, motivations, and barriers. Computers & Education, 110, 35–50. https://doi.org/10.1016/j.compedu.2017.03.003.
Shen, D., Cho, M.-H., Tsai, C.-L., & Marra, R. (2013). Unpacking online learning experiences: Online learning self-efficacy and learning satisfaction. The Internet and
Higher Education, 19, 10–17. https://doi.org/10.1016/j.iheduc.2013.04.001.
Sleeter, C. E. (2001). Preparing teachers for culturally diverse schools: Research and the overwhelming presence of whiteness. Journal of Teacher Education, 52, 94–106.
https://doi.org/10.1177/0022487101052002002.
Swan, K. (2001). Virtual interaction: Design factors affecting student satisfaction and perceived learning in asynchronous online courses. Distance Education, 22,
306–331. https://doi.org/10.1080/0158791010220208.
Vermunt, J. D., & Vermetten, Y. J. (2004). Patterns in student learning: Relationships between learning strategies, conceptions of learning, and learning orientations.
Educational Psychology Review, 16, 359–384. https://doi.org/10.1007/s10648-004-0005-y.
Wang, C.-H., Shannon, D. M., & Ross, M. E. (2013). Students' characteristics, self-regulated learning, technology self-efficacy, and course outcomes in online learning.
Distance Education, 34, 302–323. https://doi.org/10.1080/01587919.2013.835779.
Watkins, D. (2000). Learning and teaching: A cross-cultural perspective. School Leadership & Management, 20, 161–173.
Willging, P. A., & Johnson, S. D. (2009). Factors that influence students' decision to dropout of online courses. Journal of Asynchronous Learning Networks, 13, 115–127.
Wolters, C. A., & Pintrich, P. R. (1998). Contextual differences in student motivation and self-regulated learning in mathematics, English, and social studies classrooms.
Instructional Science, 26, 27–47. https://doi.org/10.1023/A:1003035929216.
Wolters, C. A., Pintrich, P. R., & Karabenick, S. A. (2005). Assessing academic self-regulated learning. In K. A. Moore, & L. H. Lippman (Vol. Eds.), What do children need
to flourish? Vol. 3, (pp. 251–270). Springer.
Yukselturk, E., & Bulut, S. (2007). Predictors for student success in an online course. Journal of Educational Technology & Society, 10.
Zheng, S., Rosson, M. B., Shih, P. C., & Carroll, J. M. (2015). Understanding student motivation, behaviors and perceptions in moocs. Proceedings of the 18th ACM
conference on computer supported cooperative work & social computing (pp. 1882–1895). ACM.
Zhu, C. (2012). Student satisfaction, performance, and knowledge construction in online collaborative learning. Journal of Educational Technology & Society, 15,
127–136.
Zhu, C. (2013). The effect of cultural and school factors on the implementation of cscl. British Journal of Educational Technology, 44, 484–501. https://doi.org/10.1111/
j.1467-8535.2012.01333.x.
Zimmerman, B. J. (2002). Becoming a self-regulated learner: An overview. Theory Into Practice, 41, 64–70. https://doi.org/10.1207/s15430421tip4102_2.
Zimmerman, B. J., & Martinez-Pons, M. (1990). Student differences in self-regulated learning: Relating grade, sex, and giftedness to self-efficacy and strategy use.
Journal of Educational Psychology, 82, 51. https://doi.org/10.1037/0022-0663.82.1.51.

30